Artikel
Markov Models and Discrete-Event-Simulation. A comparison of two powerful modelling techniques for economic evaluation
Suche in Medline nach
Autoren
Veröffentlicht: | 6. September 2007 |
---|
Gliederung
Text
Objectives: Markov models are applied to a great extend to evaluate the cost-effectiveness of competing health technologies. Recently Discrete-Event-Simulation (DES) has attract attention because of its flexibility to represent realistic clinical settings and treatment processes. This paper provides an overview of the main differences of these two modelling techniques and highlights situations where these alternatives are most appropriate.
Methods: Both modelling methods are illustrated on a real world example. The following discussion will be motivated by our own modelling experiences. A review of the leading methodological publications and recent applied studies will be presented. The focus will be a detailed discussion of major modelling issues (patient history, interactions of individuals, constrained resources).
Discussion: Markov models have the potential to describe the patient pathway over extended time horizons and to incorporate risks that are ongoing over time. These pathways are described by mutually exclusive states which represent clinically or economic events. The memoryless property of the stochastic process implies that the transition from the current state is independent from previously passed states. The individuals are assumed to be independent. These characteristics can lead to difficulties which can be overcome by DES. DES is widely used to model complex production lines and other service systems. However, it is fairly uncommon in health economics and strength and weaknesses are rarely discussed. Modellers and decision makers still face a lack of guidance for the selection of the appropriate approach.
Conclusion: The detailed discussion provides guidance for choosing an appropriate model, based on requirements and modelling effort. Therefore, it supports the improvement of decision analytic models. As a consequence it helps to increase the acceptance and the impact of these models on decision making.